Compare reported results with estimates
An earnings surprise is the difference between an actual result and an estimate. This tool uses the provider’s paired EPS and revenue fields. The comparison requires consistent currency and accounting definitions; an adjusted EPS estimate should not be casually compared with a different GAAP measure.
Calculate the percentage
Surprise % = (actual − estimate) ÷ |estimate| × 100
Illustrative values of 1.25 actual and 1.10 estimated give (1.25 − 1.10) ÷ 1.10 × 100 = 13.64%, rounded. For -0.20 actual versus -0.30 estimated, the same convention gives +33.33%, although actual earnings remain negative.
If the estimate is zero, a beat/miss comparison still works but percentage surprise is undefined. If either value is missing or calculation exceeds the finite numeric range, the percentage is unavailable. Estimates near zero can dominate an arithmetic mean.
Read the sample behind the summary
The beat rate divides EPS beats by historical records with both actual and estimated EPS. Its denominator includes zero estimates. Mean surprise uses only finite percentages, so its sample count can be smaller. Future-dated rows are excluded from summary statistics. The table filter changes visible records but not the all-record summary.
The latest run ends at a missing EPS pair, a meet or an opposite result. It refers only to returned rows: this endpoint does not establish complete or consecutive fiscal quarters. The announcement date is not necessarily the fiscal period end.
What the output cannot establish
No stock-price series is fetched, so the page cannot measure earnings reactions, prove post-announcement drift or identify a profitable strategy. Repeated beats do not by themselves demonstrate conservative guidance or deliberate under-promising. Read the release, cash flow and guidance, and check the basis and timing of the estimates.
Sources: FMP data FAQ for announcement-date and provider definitions; IFRS Foundation IAS 33 for basic/diluted earnings-per-share concepts. The formula above states this calculator’s convention; other datasets can use different surprise definitions.